INDEX

Explanations

phrases related to legal or criminal activities

oai_token-act-pair · gpt-3.5-turbo

punctuation-related marks indicating strong emotional content or finality

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 earthqu

-0.87

 eleph

-0.85

oun

-0.79

 pione

-0.79

 neighb

-0.77

Þ

-0.76

 subur

-0.75

 metic

-0.75

¥ŀ

-0.75

 prosec

-0.73

POSITIVE LOGITS

Anyway

1.70

Advertisements

1.66

Secondly

1.51

Anonymous

1.48

RAW

1.46

Reply

1.42

Contents

1.41

Rated

1.41

Yeah

1.39

CONCLUS

1.38

Activations Density 0.474%

phrases related to legal or criminal activities

punctuation-related marks indicating strong emotional content or finality

No Comments

No Known Activations